Re: [PATCH v3 0/3] tests/acpi/pcihp: add unit tests for hotplug on multifunction bridges for q35

[Ani Sinha]

On Thu, 21 Oct 2021, Michael S. Tsirkin wrote:

> On Thu, Oct 21, 2021 at 07:18:43AM +0530, Ani Sinha wrote:
> >
> >
> > On Wed, Oct 20, 2021 at 2:09 PM Michael S. Tsirkin <mst@redhat.com> wrote:
> >
> >     On Thu, Oct 07, 2021 at 07:27:47PM +0530, Ani Sinha wrote:
> >     > changelist:
> >     > v3: removed "nodefaults" from the command line and rebased the 
> > patchset.
> >     > v2: incorporated some of the feedbacks from Igor.
> >     > v1 : initial RFC patch.
> >
> >     This seems to break on s390 hosts for people. Likely an
> >     endian-ness bug somewhere. Dropped for now - care tracking that down
> >     and fixing so I can pick up the test again?
> >
> >     Thanks!
> >
> >
> > So I take it this patch wasn't causing the issue since this has been merged 
> > to
> > master now?
> >
> >
>
> Yes, we knew the bug is in the tested functionality not the test of
> course. With help from Thomas I was able to fix the original bug.
> The fix with more detail is here:
>
> commit 0e464f7d993113119f0fd17b890831440734ce15
> Author: Michael S. Tsirkin <mst@redhat.com>
> Date:   Wed Oct 20 05:48:54 2021 -0400
>
>     pci: fix PCI resource reserve capability on BE

BUG is quite old, since the time the function was actually written -

70e1ee59bb94 ("hw/pci: introduce bridge-only vendor-specific capability to
provide some hints to firmware")

I doubt in BE systems it ever worked.